The need of accurate Test Reference Years (TRYs) for simulations has been well recognised over the years. Various methods for deriving TRYs have been developed, but their final results can be significantly different. In this paper, the major methodologies reported in literature were applied to 20-year hourly measurements of weather data from Athens, covering the period 1977 to 1996, Seventeen TRYs were produced in total. The basis to select the "best" performing TRY includes meteorological criteria (inherent in the selection process used by each method) and comparisons of results from various simulations for typical energy systems (i.e. a solar water heater, a building, a large scale solar heating system with interseasonal storage and a photovoltaic system). Based on the results of each simulation er;exercise, a scoring system was developed and applied. The best performing TRY was found to be the one produced by a modified Festa-Ratto method.